Privacy Policy
Effective Date: September 2026
1. Zero Persistent Document Storage
PaperMatrix AI prioritizes academic privacy and intellectual confidentiality. Uploaded PDF files are held entirely in ephemeral RAM buffers during extraction and analysis. Once parsing completes, buffers are instantly garbage collected. No document payloads or full texts are stored on disks or centralized databases.
2. Anonymous Quota Tracking
To enforce the 3 daily free synthesis credits, we generate a one-way hardware hash via client-side fingerprinting. This hash is purely anonymous, non-reversible, and decoupled from any personal identifying information (PII) such as email, IP address, or names.
3. Third-Party Inference
Extracted structural chunks are transmitted securely over HTTPS to Groq inference hardware for low-latency synthesis. Zero user queries are used to train foundational AI models.